A Hybrid QNN-LSTM Framework for Adaptive Stock Trend Prediction Under Volatile Market Conditions

Abstract

The highly nonlinear, dynamic and volatile nature of financial temporal data makes financial stock markets difficult to predict. Traditional intelligent computational models, especially ML and DL, often cannot model the sudden market changes and long-term patterns accurately in the unstable market conditions. To address these restrictions, this paper introduces a hybrid Quantum Neural Network and Long Short-Term Memory (QNN-LSTM) framework for adaptive standard trend prediction under volatile market conditions. The proposed model takes advantage of the nonlinear feature extraction capability of QNN and the temporal learning strength of LSTM networks to improve the forecasting accuracy and enhance the prediction robustness. We perform an experimental analysis using financial time series from Yahoo Finance and NASDAQ archive. The dataset includes standard price indicators like open, close, high, low and the respective volume data as key financial indicators. The first stage is handling missing values with pre-processing and normalization methods for reducing data abnormalities. The generated data are fed into the proposed QNN-LSTM architecture for trend forecasting and volatility-aware prediction. The proposed framework is evaluated using standard performance metrics like RMSE, MAE, MAPE and prediction accuracy. The experimental results show that the QNN LSTM architecture with better prediction capability and adaptability is better than the traditional LSTM and existing deep learning methods, especially in high market volatility periods. The methodology keeps a well optimized and intelligent structure of the solution for next generation financial forecasting applications and focuses the capability of hybrid Quantum-DL methods in financial analytics.